Lymphoid Organs and Immune Traffic

Drain the thoracic duct of a rat for a few days and the lymph flowing out of it runs short of lymphocytes. If lymph nodes were making those cells, drainage should not have used them up. This lesson follows a naive lymphocyte (one that has not yet met the antigen it recognises) around the loop that explains the result, and puts numbers on how long it stays at each stop.

Lymphocytes go around a loop, not out of a factory

Lymphocytes are made in the primary lymphoid organs. In humans and mice, B cells develop in the bone marrow from blood-forming stem cells (stem cells), and T cells mature in the thymus from marrow-derived precursors. The secondary lymphoid organs (lymph nodes, the spleen, and mucosal tissue such as the tonsils and the Peyer’s patches of the gut wall) host mature lymphocytes and are where those cells meet antigen. How the thymus’s role was found, by Miller’s neonatal thymectomy in mice and by Good’s laboratory in rabbits, belongs to adaptive immunity. Two of Miller’s controls carry the logic: sham-operated mice had the same surgery without losing the thymus, and adult mice thymectomized, irradiated and given marrow could not recover immune function, unlike sham-operated controls given the same irradiation and marrow (Cross, Leuchars and Miller, 1964).

In the 1950s the usual view was that the lymphocytes in lymph were newly made in the nodes upstream. James Gowans tested it in rats fitted with a chronic cannula in the thoracic duct, the large lymph vessel that empties into the veins at the base of the neck (capillary exchange and lymph covers how lymph forms and moves). The cannula let him collect lymph for days.

The two views made different predictions. If nodes made the cells and released them once, duct output should have held roughly steady while production continued. If the same cells were going around, removing them should empty the loop. Output fell progressively. When the collected cells were returned by vein, output recovered, and the returned cells were later recovered in duct lymph.

That ruled in recirculation from blood, through lymphoid tissue, into lymph and back to blood, and ruled out the nodes as the main source of duct lymphocytes in the rat. It did not show which lymphocytes recirculate, how fast each kind does, or what molecules let them cross a vessel wall, and it gave no human values. The duct also samples only part of the pool: in the rat model fitted below it is assumed to collect lymph from the mesenteric nodes and about half of the output of the subcutaneous nodes.

Gowans’s 1964 papers asked where the cells leave the blood. Radioactively labelled lymphocytes injected by vein into rats moved quickly into lymph nodes across the walls of high endothelial venules (HEVs), small venules lined by plump, almost cuboidal endothelial cells instead of flat ones. Thome had described their shape in 1898; for decades their role went unrecognised. Gowans concluded that HEVs are the site of large-scale migration of lymphocytes from blood into the secondary organs.

A node samples lymph and blood; the spleen samples blood alone

A lymph node is a bean-shaped filter on a lymph vessel. Afferent lymphatics enter its convex surface and empty into the subcapsular sinus, a space under the capsule. Inside lie B-cell follicles in the outer cortex, a T-cell zone (the paracortex) where the HEVs sit, and medullary cords and sinuses. One efferent lymphatic, sometimes two, leaves at the hilum, beside the artery and vein.

Think of the node as a hotel with two entrances, afferent lymph from tissue and HEVs from blood, and one exit, the efferent lymphatic. The blood entrance is much the busier. At rest, afferent lymph has only about 5 to 10 percent of the cell concentration of efferent lymph, and cannulation studies in large animals found that more than 90 percent of the lymphocytes in efferent lymph had entered through HEVs.

Antigen arrives in afferent lymph, free or carried by cells, and two routes matter most here. Small soluble molecules travel from the subcapsular sinus toward the HEVs through conduits, fine tubes of reticular fibres wrapped in fibroblastic reticular cells; in mice, resident dendritic cells in the T-cell zone pick up soluble antigen this way before migrating dendritic cells arrive from the tissue. Steinman and Cohn described and named dendritic cells in 1973, in mouse spleen, lymph node and Peyer’s patch, where they were only 0.1 to 1.6 percent of nucleated cells. Their trip from skin to the draining node depends on the chemokine receptor CCR7: in CCR7-deficient mice, activated skin dendritic cells fail to reach the nodes.

The spleen, the largest secondary lymphoid organ, has no afferent lymphatics and no HEVs, so it samples blood, not lymph. Arterial blood brings both antigen and lymphocytes. Its white pulp holds T-cell and B-cell zones around arterioles, and its red pulp removes old red cells (blood cells and hemostasis). Intravital imaging in mouse spleen showed T cells leaving arterial blood into the red pulp, latching onto perivascular stromal cells without needing CCR7, then moving into the T-cell zones, which does need CCR7.

Getting in on chemokines, getting out on S1P

Entry through an HEV in a peripheral lymph node is a sequence. L-selectin (CD62L) on the lymphocyte binds sulfated mucin-like glycoproteins on the HEV, which tethers a fast-moving cell and sets it rolling. L-selectin is active all the time. The integrin LFA-1 is not: it stays in a low-affinity shape until a chemokine (CCL21, CCL19 or CXCL13) signals through its receptor, CCR7 for CCL21 and CCL19. The signal changes LFA-1’s shape so it binds ICAM-1 and ICAM-2 on the endothelium, and the cell stops, then crosses the wall. HEVs in the Peyer’s patches of the gut also use a second pair: the integrin alpha4beta7 on the lymphocyte binds MAdCAM-1 on the venule. In mice the HEVs make CCL21 themselves; in humans they do not. The HEVs also depend on lymph: tying off a node’s afferent lymphatics makes them lose their specialised features.

The molecules were found in steps. Stamper and Woodruff in 1976 laid rat and mouse lymphocytes over frozen sections of lymph node and found that they stuck to HEVs and not to other vessels. In 1983 Gallatin, Weissman and Butcher described MEL-14, an antibody to a mouse lymphocyte molecule that appeared to be required for homing to nodes; the molecule is L-selectin. In 1994 mice lacking L-selectin had a severe reduction in lymphocytes in their peripheral nodes. Cells do not all stay equally long once inside: in mice, naive CD4 T cells cross nodes about twice as fast as naive CD8 T cells.

Exit was a later question. The lipid sphingosine 1-phosphate (S1P) is high in blood and lymph and low inside lymphoid tissue, and lymphocytes carry a G protein-coupled receptor for it, S1P1. In 2004 Mehrdad Matloubian, Charles Lo, Guy Cinamon and colleagues in Jason Cyster’s laboratory asked whether lymphocytes need it to leave. Mice lacking S1P1 in every cell do not survive embryonic development, so they gave irradiated mice blood-forming cells from S1P1-deficient fetal liver. If S1P1 is needed for exit, cells should enter lymphoid organs and pile up there.

They did. Mature T cells could not leave the thymus, so no T cells reached the periphery. S1P1-deficient T and B cells transferred into normal mice could not leave lymphoid organs, so the requirement lies in the lymphocyte, not its surroundings. The study did not give human values, show how the S1P gradient is kept, or show that every subset depends on S1P1 equally. It also found that S1P1 falls when lymphocytes are activated, which goes with retention. CD69, a surface protein that appears early after activation, causes S1P1 to be internalized and degraded, and T cells expressing it stay in the node.

Residence time is set by the exit rate

Ganusov and Auerbach (2014) fitted a compartment model to data from chromium-51-labelled thoracic duct lymphocytes injected into rats. Assume first-order (constant-fraction) entry and exit, steady state, no cell death or division, and that labelled cells behave like the rest of the pool.

mean residence time = 1 / exit rate constant

number held at steady state = cells entering per unit time x mean residence time (Little’s law)

(a) Spleen: entry from blood 0.056 per min, exit 0.007 per min. Mean residence = 1 / 0.007 per min = 142.9 min, about 143 min or 2.4 h. The fitted table gives 144 min (115 to 173).

(b) In a made-up cohort of 1,000,000 labelled cells in blood, the initial rate of entry into the spleen is 0.056 per min x 1,000,000 = 56,000 cells per min.

(c) Time in blood. The model’s entry rates from blood are 1.83, 0.41, 0.056, 0.026, 0.0106 and 0.0053 per min; the two largest are into the vessels of lung and liver, from which cells return quickly. The sum is 2.338 per min, so the mean time in blood per pass is 1 / 2.338 per min = 0.428 min = 25.7 s, about 26 s, as the model reports. Of cells leaving blood, 0.056 / 2.338 = 2.4 percent go to the spleen and 0.026 / 2.338 = 1.1 percent to the subcutaneous nodes.

(d) Where the pool sits. Keep only blood (B), spleen (S) and subcutaneous nodes (N). At steady state entry equals exit for each, so S / B = 0.056 / 0.007 = 8.0. The model treats the subcutaneous nodes as two sub-compartments in series, each with exit 0.0034 per min (a mathematical device for the delay before cells leave, not two regions of a node): N1 / B = 0.026 / 0.0034 = 7.65, the second holds the same, so N / B = 15.3. Node residence = 2 / 0.0034 per min = 588 min, about 9.8 h (table: 593 min). Blood’s share is 1 / (1 + 8.0 + 15.3) = 1 / 24.3 = 4.1 percent. The spleen receives twice the node’s share of cells leaving blood but holds about half as many, because cells leave it four times faster.

(e) Mouse nodes, made-up entry: 1,000 naive CD4 and 1,000 naive CD8 T cells enter one peripheral node per hour. With residence times of 12.2 h and 21.2 h, the node holds 1,000 per h x 12.2 h = 12,200 CD4 cells and 21,200 CD8 cells, a ratio of 1.74 from residence time alone.

Answers: 143 min; 56,000 cells per min; 26 s, 2.4 and 1.1 percent; about 4 percent in blood; 12,200 and 21,200 cells. Checks: 2.338 x 0.428 = 1.00; 0.024 x 2.338 = 0.056; 24.3 x 0.041 = 1.00. The blood share is a teaching value, since lung, liver, other organs and cell loss are left out.

Spleen exit rate 0.007 per min Node exit rate, each of the two parts 0.0034 per min Lymphocytes lack S1P1 (mouse finding): node exit rate × 0.1 (schematic, not a measured value) Node drains a site of immune challenge: node entry × 2 (schematic; the lesson says only that entry may also rise, and the main change is slower exit, so try the node exit slider)

Spleen stay = 1 / exit rate

Node stay = 2 / exit rate (two parts)

Steady-state ratio blood : spleen : nodes

Steady-state shares, blood · spleen · nodes

Shares of the cohort at 24 h, blood · spleen · nodes

In this sketch: three compartments only, blood, spleen and subcutaneous nodes (two parts in series), with first-order entry and exit, steady state, no cell death or division, and labelled cells behaving like the rest of the pool; the nodes here are the lumped rat compartment. The default rates are the lesson’s fitted values for a rat (Ganusov and Auerbach, 2014; per min): spleen entry 0.056 and exit 0.007, node entry 0.026 and exit 0.0034 for each part. The slider ranges, the factor of 0.1 for lymphocytes lacking S1P1 and the doubling of node entry are made up for the sketch. The curves over 24 h are computed in this sketch from those first-order rules; the lesson gives no time course. Lung, liver, loss and other organs are omitted, so the blood share is a teaching value, not a measurement. Because lung and liver are left out, cells leave blood in about 12 min here at the default entry rates, 1 / (0.056 + 0.026) per min, not the lesson’s 26 s per pass.

Rule to keep: mean residence time = 1 / exit rate constant, and the number held at steady state = entry rate × residence time (Little’s law), so a longer stay or a faster entry gives a compartment a larger share relative to blood.

A draining node holds on to its visitors

A node that drains a site of immune challenge changes its traffic. Migrating dendritic cells arrive in afferent lymph and settle in the T-cell zone. More lymphocytes enter from blood through the HEVs, and for a time fewer leave: in mouse studies, lymphocytes in the node express CD69, CD69 removes S1P1 from their surface, and exit slows. The node enlarges and holds more lymphocytes than its partner on the other side of the body. How much of the enlargement comes from more entry and how much from less exit is still debated. A model fitted to rat data (Ganusov and Auerbach, 2014) put it down mainly to increased entry.

Little’s law gives the direction. If entry is unchanged and the exit rate constant halves, residence time doubles and so does the number held. If entry also rises, the node holds more again. The longer stay gives the rare cells that recognise the antigen more time to find it. The shutdown is transient, and exit resumes afterwards.

Design problem

You want to measure how long naive T cells stay in a mouse lymph node and test whether S1P1 sets that time. Design an adoptive-transfer experiment, say what each control rules out, and interpret these made-up results: at the start, one node holds 10,000 labelled normal T cells and 10,000 labelled S1P1-deficient T cells; 6, 12 and 24 h later it holds 6,100, 3,700 and 1,400 normal cells, and 9,800, 9,500 and 9,100 S1P1-deficient cells.

One way to do it

Take naive T cells from normal mice and from mice in which S1P1 can be deleted in T cells after they have left the thymus (for example with an inducible Cre); fetal liver chimeras will not do, because their mature T cells never reach the periphery. Give each population a different congenic surface marker, and both a photoconvertible fluorescent protein that turns from green to red under violet light. Inject them together by vein into one normal host, so differences cannot come from the host. Prepare both donor populations the same way, and swap the two markers between repeats so the labels themselves are not the cause.

A day later, expose one peripheral node to violet light under anaesthesia. Transferred cells inside it turn red; cells arriving later stay green, so new arrivals do not blur the cohort. Harvest that node from separate groups of mice at 0, 6, 12 and 24 h and count red cells of each type by flow cytometry.

Controls: equal time-zero counts show both populations start from the same cohort, so later differences reflect exit or death, and checking the viability of each population in the node helps tell them apart; illuminating cells and checking their viability tests whether the light harms them; counting red cells in blood and in an unexposed node shows where leavers go.

Interpretation: for first-order exit, cells remaining in the node = N0 x exp(-k t). For normal cells, ln(10,000 / 3,700) = 0.99 over 12 h, so k is about 0.083 per h and mean residence about 12 h; the 6 h and 24 h points fit (10,000 x exp(-0.5) = 6,070; 10,000 x exp(-2) = 1,350). For S1P1-deficient cells, ln(10,000 / 9,100) = 0.094 over 24 h, about 0.004 per h, or roughly 250 h: exit is nearly abolished, and the small loss could be cell death. Limits: one species and one node type, a few red cells may re-enter from blood, and the result shows S1P1 is needed for exit, not that it alone sets the normal residence time.